An apparatus comprising a memory, a plurality of modules, an address translation unit and a controller. The memory may be arranged as a plurality of memory banks. Each of the plurality of modules may be configured to generate one or more addresses for accessing a particular one of the plurality of memory banks. The address translation unit may be configured to modify the one or more addresses in response to a control signal. The controller may be configured to generate the control signal in response to a computer executable instruction.
|
17. A digital signal processor configured to execute one or more computer executable instructions, said computer executable instructions comprising an instruction configured to retrieve a predetermined value from a storage location, wherein one or more addresses generated by one or more modules to access a plurality of memory banks are modified in response to said predetermined value.
10. A method for associating a plurality of memory banks with one or more modules comprising the steps of:
retrieving a predetermined value in response to a computer executable instruction;
receiving an address from said one or more modules;
modifying said address in response to said predetermined value; and
presenting said modified address to a memory comprising said plurality of memory banks.
1. An apparatus comprising:
a memory arranged as a plurality of memory banks;
a plurality of modules each configured to generate one or more addresses for accessing a particular one of said plurality of memory banks;
an address translation unit configured to modify said one or more addresses in response to a control signal; and
a controller configured to generate said control signal in response to a computer executable instruction.
3. The apparatus according to
5. The apparatus according to
6. The apparatus according to
8. The apparatus according to
9. The apparatus according to
11. The method according to
said computer executable instruction is configured to retrieve said predetermined value from a storage location.
12. The method according to
13. The method according to
storing said predetermined value in said storage location prior to executing said computer executable instruction.
14. The method according to
storing a different predetermined value for each of a plurality of time slots.
15. The method according to
generating a sum of said predetermined value and each of said one or more addresses; and
applying a predetermined modulus to said sum.
16. The method according to
18. The digital signal processor according to
19. The digital signal processor according to
20. The digital signal processor according to
|
The present invention relates to a method and/or architecture for performing memory bank swapping generally and, more particularly, to a method and/or architecture for digital signal processing (DSP) memory bank rotation.
Referring to
Conventionally, DSP operations occur in a pipeline fashion. The DSP operations are divided into N pipeline stages, with each module 14a-14n working on one pipeline stage at one time. For example, a module 14j accesses a memory bank 12i in one particular time slot and in the next time slot, the module 14j accesses a memory bank 12(i+1) and so on.
Each module 14a-14n (i.e., the software or firmware) needs to keep track of the current time slot and program the memory address (i.e., bank number) accordingly. Keeping track of the current time slot and programming the memory address increases software overhead and can degrade system performance.
The present invention concerns an apparatus comprising a memory, a plurality of modules, an address translation unit and a controller. The memory may be arranged as a plurality of memory banks. Each of the plurality of modules may be configured to generate one or more addresses for accessing a particular one of the plurality of memory banks. The address translation unit may be configured to modify the one or more addresses in response to a control signal. The controller may be configured to generate the control signal in response to a computer executable instruction.
The objects, features and advantages of the present invention include providing a method and/or architecture for DSP memory bank rotation that may (i) synchronize DSP operations with one instruction, (ii) eliminate interrupts and the associated overhead, (iii) allow simultaneous access, (iv) widen the access window for direct memory access (DMA), and/or (v) minimize impact on synchronous dynamic random access memory (SDRAM) latency.
These and other objects, features and advantages of the present invention will be apparent from the following detailed description and the appended claims and drawings in which:
Referring to
Each of the modules 102a-102n may comprise an address bus 110a-110n that may be configured to generate (present) addresses for accessing a respective memory bank 112a-112n of the memory 106. Each of the busses 110a-110n may communicate a respective address (e.g., ADDR—0-ADDR_N) to the address translation block 104. The address translation block 104 may be configured to generate translated addresses (e.g., ROT_ADDR—0-ROT_ADDR_N) in response to the addresses ADDR—0-ADDR_N and a signal (e.g., ROTATE). In one example, the address translation block 104 may be configured to perform a modular addition between each of the addresses ADDR—0-ADDR_N and the signal ROTATE. However, other logical and/or arithmetic operations may be implemented accordingly to meet the design criteria of a particular application. The addresses ROT_ADDR—0-ROT_ADDR_N may be presented to the memory 106.
The signal ROTATE may be implemented, in one example, as a predetermined rotation value. By modifying the addresses ADDR—0-ADDR_N based on the signal ROTATE, each of the addresses ADDR—0-ADDR_N generally access a different one of the memory banks 112a-112n for a different value of the signal ROTATE. By changing the value of the signal ROTATE, the system 100 may rotate the memory banks 112a-112n among the modules 102a-102n without the addresses ADDR—0-ADDR_N being changed. Therefore, the signal ROTATE may determine which of the banks 112a-112n is accessed by each of the modules 102a-102n during a particular time slot.
The controller 108 may be configured to generate the signal ROTATE in response to a computer executable instruction (e.g., a new processor machine instruction swapDSP). For example, the controller 108 may be configured to present a predetermined value stored in a register 109 to the address translation unit 104 in response to the computer executable instruction. However, the predetermined value may also be stored in memory. The predetermined value may be implemented to represent a rotation amount for swapping the memory banks 112a-112n from one time slot to a subsequent time slot.
The system 100 may be configured to modify the addresses ADDR—0-ADDR_N in response to the signal ROTATE in order to swap (rotate) the memory banks 112a-112n accessed by each of the modules 102a-102n. In general, the circuit 104 is configured to modify a portion of the addresses ADDR—0-ADDR_N that determines which of the banks 112a-112n each module accesses. As the signal ROTATE changes with each time slot, a different bank 112a-112n of the memory 106 is accessed by a particular module 102a-102n while the bank determining portion of the respective addresses ADDR—0-ADDR_N generally remains unchanged from one time slot to another.
In one example, the rotation amount may be provided by the new processor machine instruction swapDSP. The computer executable instruction may be implemented, in one example, as follows:
The software design of the system 100 may be simplified with the new instruction. For example, the swapDSP instruction may specify how the addresses ADDR—0-ADDR_N may be translated to a physical bank number. In general, when the new instruction is implemented, the software of each module 102a-102n may operate without keeping track of the current time slot. In each time slot, the bank address of a particular module generally does not change. For example, a module 102j may continue using a memory bank address A while a module j+1 continues to use a memory bank address B at every time slot. At the beginning of each time slot, the computer executable instruction swapDSP may be executed to map each bank address of the modules 102a-102n to an actual bank of the memory 106. In an example application with 4 memory banks and 4 modules, the actual memory bank may be related to the module as shown in the following TABLE 1:
TABLE 1
time slot
module 0
module 1
module 2
module 3
0
bank 0
bank 1
bank 2
bank 3
1
bank 1
bank 2
bank 3
bank 0
2
bank 2
bank 3
bank 0
bank 1
3
bank 3
bank 0
bank 1
bank 2
4
bank 0
bank 1
bank 2
bank 3
5
bank 1
bank 2
bank 3
bank 0
Referring to
Referring to
Referring to
Referring to
The function performed by the flow diagram 200 of
The present invention may also be implemented by the preparation of ASICs, FPGAs, or by interconnecting an appropriate network of conventional component circuits, as is described herein, modifications of which will be readily apparent to those skilled in the art(s).
The present invention thus may also include a computer product which may be a storage medium including instructions which can be used to program a computer to perform a process in accordance with the present invention. The storage medium can include, but is not limited to, any type of disk including floppy disk, optical disk, CD-ROM, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, Flash memory, magnetic or optical cards, or any type of media suitable for storing electronic instructions.
As used herein, the term “simultaneous” is meant to describe events that share some common time period but the term is not meant to be limited to events that begin at the same point in time, end at the same point in time, or have the same duration.
While the invention has been particularly shown and described with reference to the preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made without departing from the spirit and scope of the invention.
Zhang, Fan, Leung, Ho-Ming, Koe, Wern-Yan, Rangam, Kasturiranga N., Balasubramanian, Venkatesh
Patent | Priority | Assignee | Title |
11232027, | Mar 01 2016 | Sony Corporation | Memory control apparatus, memory apparatus, information processing system, and memory control method |
7779198, | Nov 23 2004 | Efficient Memory Technology | Method and apparatus of multiple abbreviations of interleaved addressing of paged memories |
8074010, | Nov 23 2004 | Efficient Memory Technology | Intelligent memory banks for storing vectors |
8190809, | Nov 23 2004 | Efficient Memory Technology | Shunted interleave for accessing plural memory banks, particularly those having partially accessed cells containing data for cache lines |
Patent | Priority | Assignee | Title |
4493027, | May 22 1981 | Data General Corporation | Method of performing a call operation in a digital data processing system having microcode call and return operations |
4633434, | Apr 02 1984 | Sperry Corporation | High performance storage unit |
4888687, | May 04 1987 | Bankers Trust Company | Memory control system |
4894770, | Jun 01 1987 | Massachusetts Institute of Technology | Set associative memory |
4930066, | Oct 15 1985 | Agency of Industrial Science and Technology | Multiport memory system |
5269010, | Aug 31 1990 | Advanced Micro Devices, Inc. | Memory control for use in a memory system incorporating a plurality of memory banks |
5835963, | Sep 09 1994 | Hitachi, LTD | Processor with an addressable address translation buffer operative in associative and non-associative modes |
5924111, | Oct 17 1995 | MEDIATEK INC | Method and system for interleaving data in multiple memory bank partitions |
6215719, | Dec 22 1998 | RPX Corporation | Memory device having line address counter for making next line active while current line is processed |
6289429, | Mar 11 1996 | SOCIONEXT INC | Accessing multiple memories using address conversion among multiple addresses |
6356991, | Dec 31 1997 | Unisys Corporation | Programmable address translation system |
6381668, | Mar 21 1997 | TWITTER, INC | Address mapping for system memory |
6553478, | May 03 1999 | STMICROELECTRONICS S A | Computer memory access |
20030140305, |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
May 20 2002 | BALASUBRAMANIAN, VENKATESH | LSI Logic Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012930 | /0257 | |
May 20 2002 | RANGAM, KASTURIRANGA | LSI Logic Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012930 | /0257 | |
May 20 2002 | ZHANG, JENNIFER F | LSI Logic Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012930 | /0257 | |
May 20 2002 | KOE, WERN-YAN | LSI Logic Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012930 | /0257 | |
May 20 2002 | LEUNG, HO-MING | LSI Logic Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 012930 | /0257 | |
May 22 2002 | LSI Logic Corporation | (assignment on the face of the patent) | / | |||
Apr 06 2007 | LSI Logic Corporation | LSI Corporation | CHANGE OF NAME SEE DOCUMENT FOR DETAILS | 033102 | /0270 | |
May 06 2014 | Agere Systems LLC | DEUTSCHE BANK AG NEW YORK BRANCH, AS COLLATERAL AGENT | PATENT SECURITY AGREEMENT | 032856 | /0031 | |
May 06 2014 | LSI Corporation | DEUTSCHE BANK AG NEW YORK BRANCH, AS COLLATERAL AGENT | PATENT SECURITY AGREEMENT | 032856 | /0031 | |
Aug 14 2014 | LSI Corporation | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 035390 | /0388 | |
Feb 01 2016 | DEUTSCHE BANK AG NEW YORK BRANCH, AS COLLATERAL AGENT | Agere Systems LLC | TERMINATION AND RELEASE OF SECURITY INTEREST IN PATENT RIGHTS RELEASES RF 032856-0031 | 037684 | /0039 | |
Feb 01 2016 | DEUTSCHE BANK AG NEW YORK BRANCH, AS COLLATERAL AGENT | LSI Corporation | TERMINATION AND RELEASE OF SECURITY INTEREST IN PATENT RIGHTS RELEASES RF 032856-0031 | 037684 | /0039 | |
Feb 01 2016 | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | BANK OF AMERICA, N A , AS COLLATERAL AGENT | PATENT SECURITY AGREEMENT | 037808 | /0001 | |
Jan 19 2017 | BANK OF AMERICA, N A , AS COLLATERAL AGENT | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | TERMINATION AND RELEASE OF SECURITY INTEREST IN PATENTS | 041710 | /0001 |
Date | Maintenance Fee Events |
Apr 01 2008 | ASPN: Payor Number Assigned. |
Nov 17 2008 | M1551: Payment of Maintenance Fee, 4th Year, Large Entity. |
Sep 28 2012 | M1552: Payment of Maintenance Fee, 8th Year, Large Entity. |
Dec 23 2016 | REM: Maintenance Fee Reminder Mailed. |
May 17 2017 | EXP: Patent Expired for Failure to Pay Maintenance Fees. |
Date | Maintenance Schedule |
May 17 2008 | 4 years fee payment window open |
Nov 17 2008 | 6 months grace period start (w surcharge) |
May 17 2009 | patent expiry (for year 4) |
May 17 2011 | 2 years to revive unintentionally abandoned end. (for year 4) |
May 17 2012 | 8 years fee payment window open |
Nov 17 2012 | 6 months grace period start (w surcharge) |
May 17 2013 | patent expiry (for year 8) |
May 17 2015 | 2 years to revive unintentionally abandoned end. (for year 8) |
May 17 2016 | 12 years fee payment window open |
Nov 17 2016 | 6 months grace period start (w surcharge) |
May 17 2017 | patent expiry (for year 12) |
May 17 2019 | 2 years to revive unintentionally abandoned end. (for year 12) |